BACKGROUND Meta-analyses have indicated that preoperative mechanical bowel preparation (MBP) confers no clear benefit and may indeed be harmful for patients with colorectal cancer. The effects of bowel preparation on longer-term outcomes have not been reported. The aim was to compare long-term survival and surgical complications in patients who did or did not receive MBP before surgery for colonic cancer. METHODS This was a retrospective cohort study of all patients undergoing potentially curative surgery for colonic cancer after routine hospital admission in the West of Scotland between January 2000 and December 2005. Clinical audit data were linked to cancer registrations and death certificates. Kaplan-Meier and Cox proportional hazards models were used to explore determinants of survival. RESULTS A total of 1730 patients underwent potentially curative surgery for colonic cancer, of whom 886 (51·2 per cent) were men. The mean(s.d.) age was 69·7(10·6) years. Some 1460 patients (84·4 per cent) received MBP. Median follow-up was 3·5 (range 0·1-6·7) years. There were no statistically significant differences in 30-day postoperative complication rates between groups. The unadjusted hazard ratio (HR) for death from all causes for patients treated with MBP (versus no MBP) was 0·72 (95 per cent confidence interval 0·57 to 0·91). Multivariable analysis with adjustment for age, sex, socioeconomic circumstances, disease stage and presentation for surgery showed that MBP had no independent effect on all-cause mortality (HR 0·85, 0·67 to 1·10). CONCLUSION Neither postoperative complications nor long-term survival are improved by MBP before colonic cancer surgery.